Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
0.00% |
0 / 180 |
|
0.00% |
0 / 13 |
CRAP | |
0.00% |
0 / 1 |
| TextHandler | |
0.00% |
0 / 180 |
|
0.00% |
0 / 13 |
2756 | |
0.00% |
0 / 1 |
| __construct | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
2 | |||
| getTracks | |
0.00% |
0 / 7 |
|
0.00% |
0 / 1 |
20 | |||
| getTimedTextNamespace | |
0.00% |
0 / 25 |
|
0.00% |
0 / 1 |
110 | |||
| getTextPagesFromDb | |
0.00% |
0 / 20 |
|
0.00% |
0 / 1 |
12 | |||
| getRemoteTextPagesQuery | |
0.00% |
0 / 18 |
|
0.00% |
0 / 1 |
6 | |||
| getRemoteTextSources | |
0.00% |
0 / 24 |
|
0.00% |
0 / 1 |
20 | |||
| getForeignDbTextSources | |
0.00% |
0 / 4 |
|
0.00% |
0 / 1 |
6 | |||
| getLocalDbTextSources | |
0.00% |
0 / 4 |
|
0.00% |
0 / 1 |
6 | |||
| getTextTracksFromRows | |
0.00% |
0 / 25 |
|
0.00% |
0 / 1 |
30 | |||
| getTextTracksFromData | |
0.00% |
0 / 8 |
|
0.00% |
0 / 1 |
42 | |||
| getContentType | |
0.00% |
0 / 5 |
|
0.00% |
0 / 1 |
12 | |||
| getFullURL | |
0.00% |
0 / 18 |
|
0.00% |
0 / 1 |
6 | |||
| convertSubtitles | |
0.00% |
0 / 20 |
|
0.00% |
0 / 1 |
72 | |||
| 1 | <?php |
| 2 | /** |
| 3 | * Timed Text handling for mediaWiki |
| 4 | * |
| 5 | * Timed text support is presently fairly limited. Unlike Ogg and WebM handlers, |
| 6 | * timed text does not extend the TimedMediaHandler class. |
| 7 | * |
| 8 | * TODO On "new" timedtext language save purge all pages where file exists |
| 9 | */ |
| 10 | |
| 11 | namespace MediaWiki\TimedMediaHandler\Handlers\TextHandler; |
| 12 | |
| 13 | use Exception; |
| 14 | use File; |
| 15 | use ForeignDBFile; |
| 16 | use IForeignRepoWithDB; |
| 17 | use IForeignRepoWithMWApi; |
| 18 | use LocalRepo; |
| 19 | use MediaWiki\MediaWikiServices; |
| 20 | use MediaWiki\TimedMediaHandler\TimedText\ParseError; |
| 21 | use MediaWiki\TimedMediaHandler\TimedText\SrtReader; |
| 22 | use MediaWiki\TimedMediaHandler\TimedText\SrtWriter; |
| 23 | use MediaWiki\TimedMediaHandler\TimedText\VttWriter; |
| 24 | use MediaWiki\TimedMediaHandler\TimedTextPage; |
| 25 | use MediaWiki\Title\Title; |
| 26 | use RuntimeException; |
| 27 | use Wikimedia\Rdbms\IExpression; |
| 28 | use Wikimedia\Rdbms\IResultWrapper; |
| 29 | use Wikimedia\Rdbms\LikeValue; |
| 30 | |
| 31 | class TextHandler { |
| 32 | /** @var int|null lazy init remote Namespace number */ |
| 33 | public $remoteNs; |
| 34 | /** @var string|null lazy init remote Namespace name */ |
| 35 | public $remoteNsName; |
| 36 | |
| 37 | /** |
| 38 | * @var File |
| 39 | */ |
| 40 | protected $file; |
| 41 | |
| 42 | /** |
| 43 | * @var array of string keys for subtitle formats |
| 44 | */ |
| 45 | protected $formats; |
| 46 | |
| 47 | /** |
| 48 | * @param File $file |
| 49 | * @param array $formats |
| 50 | */ |
| 51 | public function __construct( |
| 52 | $file, |
| 53 | $formats = [ TimedTextPage::VTT_SUBTITLE_FORMAT, TimedTextPage::SRT_SUBTITLE_FORMAT ] |
| 54 | ) { |
| 55 | $this->file = $file; |
| 56 | $this->formats = $formats; |
| 57 | } |
| 58 | |
| 59 | /** |
| 60 | * Get the timed text tracks elements as an associative array |
| 61 | * @return array[] |
| 62 | */ |
| 63 | public function getTracks() { |
| 64 | if ( $this->file->isLocal() ) { |
| 65 | return $this->getLocalDbTextSources(); |
| 66 | } |
| 67 | if ( $this->file instanceof ForeignDBFile ) { |
| 68 | return $this->getForeignDbTextSources(); |
| 69 | } |
| 70 | if ( $this->file->getRepo() instanceof IForeignRepoWithMWApi ) { |
| 71 | return $this->getRemoteTextSources( $this->file ); |
| 72 | } |
| 73 | return []; |
| 74 | } |
| 75 | |
| 76 | /** |
| 77 | * @return false|int |
| 78 | */ |
| 79 | public function getTimedTextNamespace() { |
| 80 | $repo = $this->file->getRepo(); |
| 81 | |
| 82 | if ( $this->file->isLocal() ) { |
| 83 | return NS_TIMEDTEXT; |
| 84 | } |
| 85 | if ( $repo instanceof IForeignRepoWithDB ) { |
| 86 | $config = MediaWikiServices::getInstance()->getMainConfig(); |
| 87 | $timedTextForeignNamespaces = $config->get( 'TimedTextForeignNamespaces' ); |
| 88 | $wikiID = $repo->getReplicaDB()->getDomainID(); |
| 89 | // if failed to get namespace via IForeignRepoWithDB, return NS_TIMEDTEXT |
| 90 | return $timedTextForeignNamespaces[$wikiID] ?? NS_TIMEDTEXT; |
| 91 | } |
| 92 | if ( $repo instanceof IForeignRepoWithMWApi ) { |
| 93 | if ( $this->remoteNs !== null ) { |
| 94 | return $this->remoteNs; |
| 95 | } |
| 96 | |
| 97 | // Get the namespace data from the image api repo: |
| 98 | // fetchImageQuery query caches results |
| 99 | $data = $repo->fetchImageQuery( [ |
| 100 | 'meta' => 'siteinfo', |
| 101 | 'siprop' => 'namespaces' |
| 102 | ] ); |
| 103 | |
| 104 | if ( isset( $data['query']['namespaces'] ) ) { |
| 105 | // get the ~last~ timed text namespace defined |
| 106 | foreach ( $data['query']['namespaces'] as $ns ) { |
| 107 | if ( isset( $ns['canonical'] ) && $ns['canonical'] === 'TimedText' ) { |
| 108 | $this->remoteNs = $ns['id']; |
| 109 | $this->remoteNsName = $ns['*']; |
| 110 | wfDebug( "Discovered remoteNs: $this->remoteNs and name: $this->remoteNsName \n" ); |
| 111 | break; |
| 112 | } |
| 113 | } |
| 114 | } |
| 115 | |
| 116 | // Return the remote Ns, if found |
| 117 | if ( $this->remoteNs !== null ) { |
| 118 | return $this->remoteNs; |
| 119 | } |
| 120 | } |
| 121 | |
| 122 | return false; |
| 123 | } |
| 124 | |
| 125 | /** |
| 126 | * Retrieve a list of TimedText pages in the database that start with |
| 127 | * the name of the file associated with this handler. |
| 128 | * |
| 129 | * If the file is on a foreign repo, will query the ForeignDb |
| 130 | * |
| 131 | * @return IResultWrapper|false |
| 132 | */ |
| 133 | private function getTextPagesFromDb() { |
| 134 | $ns = $this->getTimedTextNamespace(); |
| 135 | if ( $ns === false ) { |
| 136 | wfDebug( 'Repo: ' . $this->file->getRepoName() . " does not have a TimedText namespace \n" ); |
| 137 | // No timed text namespace, don't try to look up timed text tracks |
| 138 | return false; |
| 139 | } |
| 140 | |
| 141 | $repo = $this->file->getRepo(); |
| 142 | if ( $repo instanceof LocalRepo ) { |
| 143 | $dbr = $repo->getReplicaDB(); |
| 144 | $prefix = $this->file->getTitle()->getDBkey(); |
| 145 | return $dbr->newSelectQueryBuilder() |
| 146 | ->select( [ 'page_namespace', 'page_title' ] ) |
| 147 | ->from( 'page' ) |
| 148 | ->where( [ |
| 149 | 'page_namespace' => $ns, |
| 150 | $dbr->expr( 'page_title', IExpression::LIKE, new LikeValue( $prefix, $dbr->anyString() ) ), |
| 151 | ] ) |
| 152 | ->limit( 300 ) |
| 153 | ->orderBy( 'page_title' ) |
| 154 | ->caller( __METHOD__ ) |
| 155 | ->fetchResultSet(); |
| 156 | } |
| 157 | |
| 158 | return false; |
| 159 | } |
| 160 | |
| 161 | /** |
| 162 | * Build the api query to find TimedText pages belonging to a remote file |
| 163 | * @return array|false |
| 164 | */ |
| 165 | public function getRemoteTextPagesQuery() { |
| 166 | $ns = $this->getTimedTextNamespace(); |
| 167 | if ( $ns === false ) { |
| 168 | wfDebug( 'Repo: ' . $this->file->getRepoName() . " does not have a TimedText namespace \n" ); |
| 169 | // No timed text namespace, don't try to look up timed text tracks |
| 170 | return false; |
| 171 | } |
| 172 | $canonicalTitle = Title::makeName( |
| 173 | $this->file->getTitle()->getNamespace(), |
| 174 | $this->file->getTitle()->getDbKey(), |
| 175 | '', |
| 176 | '', |
| 177 | true |
| 178 | ); |
| 179 | return [ |
| 180 | 'action' => 'query', |
| 181 | // For a remote wiki, we need to always use canonical namespace names |
| 182 | 'titles' => $canonicalTitle, |
| 183 | 'prop' => 'videoinfo', |
| 184 | 'viprop' => 'timedtext', |
| 185 | 'formatversion' => '2', |
| 186 | ]; |
| 187 | } |
| 188 | |
| 189 | /** |
| 190 | * Retrieve the text sources belonging to a remote file |
| 191 | * @param File $file The File's repo must implement IForeignRepoWithMWApi |
| 192 | * @return array[] |
| 193 | */ |
| 194 | private function getRemoteTextSources( File $file ) { |
| 195 | $regenerator = function () use ( $file ) { |
| 196 | /** @var IForeignRepoWithMWApi $repo */ |
| 197 | $repo = $file->getRepo(); |
| 198 | '@phan-var IForeignRepoWithMWApi $repo'; |
| 199 | wfDebug( "Get text tracks from remote api \n" ); |
| 200 | $query = $this->getRemoteTextPagesQuery(); |
| 201 | // Error in getting timed text namespace return empty array; |
| 202 | if ( $query === false || !( $repo instanceof IForeignRepoWithMWApi ) ) { |
| 203 | return []; |
| 204 | } |
| 205 | |
| 206 | $data = $repo->fetchImageQuery( $query ); |
| 207 | |
| 208 | return $this->getTextTracksFromData( $data ); |
| 209 | }; |
| 210 | |
| 211 | $repoInfo = $file->getRepo()->getInfo(); |
| 212 | $cacheTTL = $repoInfo['descriptionCacheExpiry'] ?? 0; |
| 213 | |
| 214 | if ( $cacheTTL > 0 ) { |
| 215 | $cache = MediaWikiServices::getInstance()->getMainWANObjectCache(); |
| 216 | $textTracks = $cache->getWithSetCallback( |
| 217 | $cache->makeKey( |
| 218 | 'RemoteTextTracks-url', |
| 219 | $this->file->getRepoName(), |
| 220 | $this->file->getName() |
| 221 | ), |
| 222 | $cacheTTL, |
| 223 | $regenerator |
| 224 | ); |
| 225 | } else { |
| 226 | $textTracks = $regenerator(); |
| 227 | } |
| 228 | |
| 229 | return $textTracks; |
| 230 | } |
| 231 | |
| 232 | /** |
| 233 | * Retrieve the text sources belonging to a foreign db accessible file |
| 234 | * @return array[] |
| 235 | */ |
| 236 | public function getForeignDbTextSources() { |
| 237 | $data = $this->getTextPagesFromDb(); |
| 238 | if ( $data !== false ) { |
| 239 | return $this->getTextTracksFromRows( $data ); |
| 240 | } |
| 241 | return []; |
| 242 | } |
| 243 | |
| 244 | /** |
| 245 | * Retrieve the text sources belonging to a local file |
| 246 | * @return array[] |
| 247 | */ |
| 248 | public function getLocalDbTextSources() { |
| 249 | $data = $this->getTextPagesFromDb(); |
| 250 | if ( $data !== false ) { |
| 251 | return $this->getTextTracksFromRows( $data ); |
| 252 | } |
| 253 | return []; |
| 254 | } |
| 255 | |
| 256 | /** |
| 257 | * Build an array of track information using a Database result |
| 258 | * Handles both local and foreign Db results |
| 259 | * |
| 260 | * @param IResultWrapper $data Database result with page titles |
| 261 | * @return array[] |
| 262 | */ |
| 263 | public function getTextTracksFromRows( IResultWrapper $data ) { |
| 264 | $textTracks = []; |
| 265 | |
| 266 | $services = MediaWikiServices::getInstance(); |
| 267 | $langNames = $services->getLanguageNameUtils()->getLanguageNames(); |
| 268 | $languageFactory = $services->getLanguageFactory(); |
| 269 | |
| 270 | foreach ( $data as $row ) { |
| 271 | $titleParts = explode( '.', $row->page_title ); |
| 272 | if ( count( $titleParts ) >= 3 ) { |
| 273 | $timedTextExtension = array_pop( $titleParts ); |
| 274 | $languageKey = array_pop( $titleParts ); |
| 275 | } else { |
| 276 | continue; |
| 277 | } |
| 278 | // If there is no valid language continue: |
| 279 | if ( !isset( $langNames[ $languageKey ] ) ) { |
| 280 | continue; |
| 281 | } |
| 282 | |
| 283 | $language = $languageFactory->getLanguage( $languageKey ); |
| 284 | foreach ( $this->formats as $format ) { |
| 285 | $textTracks[] = [ |
| 286 | 'src' => $this->getFullURL( $languageKey, $format ), |
| 287 | 'kind' => 'subtitles', |
| 288 | 'type' => $this->getContentType( $format ), |
| 289 | 'srclang' => $language->getHtmlCode(), |
| 290 | 'dir' => $language->getDir(), |
| 291 | 'label' => wfMessage( 'timedmedia-subtitle-language', |
| 292 | $langNames[ $languageKey ], |
| 293 | $languageKey )->text() |
| 294 | ]; |
| 295 | } |
| 296 | } |
| 297 | return $textTracks; |
| 298 | } |
| 299 | |
| 300 | /** |
| 301 | * Build an array of track information using an API result |
| 302 | * @param mixed $data JSON decoded result from a query API request |
| 303 | * @return array[] |
| 304 | */ |
| 305 | public function getTextTracksFromData( $data ) { |
| 306 | $textTracks = []; |
| 307 | foreach ( $data['query']['pages'] ?? [] as $page ) { |
| 308 | foreach ( $page['videoinfo'] ?? [] as $info ) { |
| 309 | foreach ( $info['timedtext'] ?? [] as $track ) { |
| 310 | foreach ( $this->formats as $format ) { |
| 311 | if ( ( $track['type'] ?? '' ) === $this->getContentType( $format ) ) { |
| 312 | // Add validation ? |
| 313 | $textTracks[] = $track; |
| 314 | } |
| 315 | } |
| 316 | } |
| 317 | } |
| 318 | } |
| 319 | return $textTracks; |
| 320 | } |
| 321 | |
| 322 | /** |
| 323 | * @param string $timedTextExtension |
| 324 | * |
| 325 | * @return string |
| 326 | */ |
| 327 | public function getContentType( $timedTextExtension ) { |
| 328 | if ( $timedTextExtension === TimedTextPage::SRT_SUBTITLE_FORMAT ) { |
| 329 | return 'text/x-srt'; |
| 330 | } |
| 331 | if ( $timedTextExtension === TimedTextPage::VTT_SUBTITLE_FORMAT ) { |
| 332 | return 'text/vtt'; |
| 333 | } |
| 334 | return ''; |
| 335 | } |
| 336 | |
| 337 | /** |
| 338 | * Retrieve a url to the raw subtitle file |
| 339 | * Only use for local and foreignDb requests |
| 340 | * |
| 341 | * @param string $lang |
| 342 | * @param string $format |
| 343 | * @return string |
| 344 | */ |
| 345 | public function getFullURL( $lang, $format ) { |
| 346 | $title = $this->file->getTitle(); |
| 347 | // Note we need to use the canonical namespace in case this is a |
| 348 | // foreign DB repo (Wikimedia Commons style) in a different language |
| 349 | // than the current site. |
| 350 | $canonicalTitle = Title::makeName( |
| 351 | $title->getNamespace(), |
| 352 | $title->getDbKey(), |
| 353 | '', |
| 354 | '', |
| 355 | true |
| 356 | ); |
| 357 | $params = [ |
| 358 | 'action' => 'timedtext', |
| 359 | 'title' => $canonicalTitle, |
| 360 | 'lang' => $lang, |
| 361 | 'trackformat' => $format, |
| 362 | ]; |
| 363 | if ( !$this->file->isLocal() ) { |
| 364 | $params['origin'] = '*'; |
| 365 | } |
| 366 | $query = wfArrayToCgi( $params ); |
| 367 | |
| 368 | // Note: This will return false if scriptDirUrl is not set for repo. |
| 369 | return $this->file->getRepo()->makeUrl( $query, 'api' ); |
| 370 | } |
| 371 | |
| 372 | /** |
| 373 | * Convert subtitles between SubRIP (SRT) and WebVTT, laxly. |
| 374 | * |
| 375 | * @param string $from source format, one of TimedTextPage::SRT_SUBTITLE_FORMAT |
| 376 | * or TimedTextPage::VTT_SUBTITLE_FORMAT |
| 377 | * @param string $to destination format, one of TimedTextPage::SRT_SUBTITLE_FORMAT |
| 378 | * or TimedTextPage::VTT_SUBTITLE_FORMAT |
| 379 | * @param string $data source-formatted subtitles |
| 380 | * @param ParseError[] &$errors optional outparam to capture errors |
| 381 | * @return string destination-formatted subtitles |
| 382 | */ |
| 383 | public static function convertSubtitles( $from, $to, $data, &$errors = [] ) { |
| 384 | // Note that we convert even if the format is the same, to ensure |
| 385 | // data format integrity. |
| 386 | // |
| 387 | // @todo cache the conversion in memcached |
| 388 | switch ( $from ) { |
| 389 | case TimedTextPage::SRT_SUBTITLE_FORMAT: |
| 390 | $reader = new SrtReader(); |
| 391 | break; |
| 392 | case TimedTextPage::VTT_SUBTITLE_FORMAT: |
| 393 | // @todo once VttReader is implemented, use it. |
| 394 | // For now throw an exception rather than a fatal error. |
| 395 | throw new RuntimeException( 'vtt source pages are not yet supported' ); |
| 396 | default: |
| 397 | throw new RuntimeException( 'Unsupported timedtext filetype' ); |
| 398 | } |
| 399 | switch ( $to ) { |
| 400 | case TimedTextPage::SRT_SUBTITLE_FORMAT: |
| 401 | $writer = new SrtWriter(); |
| 402 | break; |
| 403 | case TimedTextPage::VTT_SUBTITLE_FORMAT: |
| 404 | $writer = new VttWriter(); |
| 405 | break; |
| 406 | default: |
| 407 | throw new RuntimeException( 'Unsupported timedtext filetype' ); |
| 408 | } |
| 409 | try { |
| 410 | $reader->read( $data ); |
| 411 | $cues = $reader->getCues(); |
| 412 | $errors = $reader->getErrors(); |
| 413 | |
| 414 | return $writer->write( $cues ); |
| 415 | } catch ( Exception $e ) { |
| 416 | throw new RuntimeException( 'Timed text track conversion failed: ' . |
| 417 | $e->getMessage() ); |
| 418 | } |
| 419 | } |
| 420 | } |